Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!mit-eddie!genrad!decvax!decwrl!nsc!nsta!instable!amos
From: amos@instable.UUCP
Newsgroups: comp.arch,comp.sys.nsc.32k
Subject: Re: Branch prediction in the 532
Message-ID: <752@instable.UUCP>
Date: Wed, 22-Apr-87 10:32:06 EST
Article-I.D.: instable.752
Posted: Wed Apr 22 10:32:06 1987
Date-Received: Fri, 24-Apr-87 02:32:25 EST
References: <324@dumbo.UUCP>
Reply-To: amos%nsta@nsc.com (Amos Shapir)
Distribution: world
Organization: National Semiconductor (Israel) Ltd. Home of the 32532
Lines: 16
Xref: utgpu comp.arch:1000 comp.sys.nsc.32k:105
Summary: Someone got it wrong
Hdate: 23 Nisan 5747

Looks like the person who wrote the brochure about the 32532 didn't
look at the spec; anyway it's certain that:

1. The whole point of branch prediction is not waiting for the condition
  codes; therefore the prediction cannot, and does not, rely on them.

2. Forward branches are predicted as not taken, backward branches are
  predicted as taken, no matter who says otherwise. (We have tried
  several methods of state machines, caches, etc. and found out that
  this simple-minded method had the best ratio of relative performance
  increase vs. cost in chip area and overhead.)
-- 
	Amos Shapir
National Semiconductor (Israel)
6 Maskit st. P.O.B. 3007, Herzlia 46104, Israel  Tel. (972)52-522261
amos%nsta@nsc.com {hplabs,pyramid,sun,decwrl} 34.48'E 32.10'N